Erkennung der Bug- und Heckwellen von Schiffen durch satellitenbasierte C-Band- und X-Band-Radarsensoren mit synthetischer Apertur
Publication date
2025-06-20
Document type
Dissertation
Author
Tings, Björn
Advisor
Referee
Soloviev, Alexander
Granting institution
Helmut-Schmidt-Universität/Universität der Bundeswehr Hamburg
Exam date
2025-04-28
Organisational unit
Publisher
Universitätsbibliothek der HSU/UniBw H
Part of the university bibliography
✅
Language
German
DDC Class
620 Ingenieurwissenschaften
Keyword
Synthetic Aperture Radar (SAR)
Maritime Situational Awareness (MSA)
Oceanography
Object recognition
Object detectability
Ship detection
Wake detection
Detectability modelling
Detectability comparison
Image processing
Pattern recognition
Machine learning
Deep learning
C-Band
X-Band
TerraSAR-X
CosmoSkymed
Sentinel-1
RADARSAT-2
Ocean surface imaging
Support vector machine
YOLO
Near Real-Time (NRT)
Abstract
Ship wakes originate from interaction of ship’s hull and the surrounding water. They are a result of multiple superimposing and interacting wave systems closely beneath and on the ocean surface.
Synthetic Aperture Radar (SAR) with C-band and X-band radar frequencies is used in maritime situational awareness systems for surveillance of ship traffic. An advantage of SAR is that no co-operation on ship side is required. Accordingly, SAR is applied complementarily to data sources, which require cooperation. Wake signatures of ships are represented in SAR imagery as complex structures consisting of multiple wake components. Wake signatures can be exploited for indirect detection of ships, if direct detection of ships is complicated or impossible due to absent or weak ship signatures. This is especially the case for small ships, like fishing boats. The goal of this work is to increase the usefulness of maritime situational awareness for surveil-lance of ship traffic. Focus is on the recognition of ship wakes in SAR imagery. The automatic detection of wakes is subject of research for decades. However, in the present work, the de-pendency of detectability from the influencing parameters is systematically analyzed for the first time using real SAR data. Theoretical foundations for this analysis are machine learning methods and the accumulated knowledge on radar backscatter properties of the ocean surface. On that Basis, the dependencies of detectability from influencing parameters are reproduced by so-called detectability models. A figure of merit for detectability and a measure for uncertainty of detectability models is developed. The new results are contrasted against scientific literature, which previously is based on simulations and/or physical deductions on ship wakes and their recognition in SAR imagery. According to the current state-of-the-art the scientific literature exhibits research gaps.
With the new approach of a systematic analysis research gaps are closed. The potential of applying wake recognition for the purpose of ship detection can now be evaluated quantitatively. By means of use cases is demonstrated that the developed detectability models can be applied to control the precision performance of wake detectors. Additionally, a new method for estimation of vessel velocity is presented, whose accuracy coincides with other published methods. Consequently, the results of this work imply three solutions for increasing the usefulness of maritime situational awareness systems.
Synthetic Aperture Radar (SAR) with C-band and X-band radar frequencies is used in maritime situational awareness systems for surveillance of ship traffic. An advantage of SAR is that no co-operation on ship side is required. Accordingly, SAR is applied complementarily to data sources, which require cooperation. Wake signatures of ships are represented in SAR imagery as complex structures consisting of multiple wake components. Wake signatures can be exploited for indirect detection of ships, if direct detection of ships is complicated or impossible due to absent or weak ship signatures. This is especially the case for small ships, like fishing boats. The goal of this work is to increase the usefulness of maritime situational awareness for surveil-lance of ship traffic. Focus is on the recognition of ship wakes in SAR imagery. The automatic detection of wakes is subject of research for decades. However, in the present work, the de-pendency of detectability from the influencing parameters is systematically analyzed for the first time using real SAR data. Theoretical foundations for this analysis are machine learning methods and the accumulated knowledge on radar backscatter properties of the ocean surface. On that Basis, the dependencies of detectability from influencing parameters are reproduced by so-called detectability models. A figure of merit for detectability and a measure for uncertainty of detectability models is developed. The new results are contrasted against scientific literature, which previously is based on simulations and/or physical deductions on ship wakes and their recognition in SAR imagery. According to the current state-of-the-art the scientific literature exhibits research gaps.
With the new approach of a systematic analysis research gaps are closed. The potential of applying wake recognition for the purpose of ship detection can now be evaluated quantitatively. By means of use cases is demonstrated that the developed detectability models can be applied to control the precision performance of wake detectors. Additionally, a new method for estimation of vessel velocity is presented, whose accuracy coincides with other published methods. Consequently, the results of this work imply three solutions for increasing the usefulness of maritime situational awareness systems.
Version
Published version
Access right on openHSU
Open access